At 03:17 UTC on a Tuesday, our on-call pager went off: Kafka consumer lag for the payments topic had hit 12,478,291 messages, with p99 processing latency spiking to 8.7 seconds. We were running Kafka 3.7.0, Kubernetes 1.32.0, KEDA 2.14.1, and Prometheus 2.50.1 — a stack we’d battle-tested for 18 months. This wasn’t a slow leak. This was a sudden, catastrophic failure that threatened to delay 40% of our daily payment processing volume. 📡 Hacker News Top Stories Right Now Ghostty is leaving GitHub (1860 points) Before GitHub (292 points) How ChatGPT serves ads (187 points) We decreased our LLM costs with Opus (49 points) Regression: malware reminder on every read still causes subagent refusals (156 points) Key Insights Kafka 3.7’s default consumer session timeout (45s) conflicts with K8s 1.32’s default pod graceful shutdown period (30s) when KEDA 2.14 scales consumers aggressively.…